Skip to main content

treeship_core/session/
receipt.rs

1//! Session Receipt composer.
2//!
3//! Builds the canonical Session Receipt JSON from session events,
4//! artifact store, and Merkle tree. The receipt is the composed
5//! package-level artifact that unifies an entire session.
6
7use serde::{Deserialize, Serialize};
8use sha2::{Sha256, Digest};
9
10use crate::merkle::{MerkleTree, InclusionProof};
11
12use super::event::SessionEvent;
13use super::graph::AgentGraph;
14use super::manifest::{
15    HostInfo, LifecycleMode, Participants, SessionManifest, SessionStatus, ToolInfo,
16};
17use super::render::RenderConfig;
18use super::side_effects::SideEffects;
19
20/// Receipt type identifier.
21pub const RECEIPT_TYPE: &str = "treeship/session-receipt/v1";
22
23/// Current receipt schema version. Receipts without this field are treated
24/// as schema "0" and verified under legacy rules (pre-v0.9.0 shape).
25pub const RECEIPT_SCHEMA_VERSION: &str = "1";
26
27// ── Top-level receipt ────────────────────────────────────────────────
28
29/// The complete Session Receipt.
30#[derive(Debug, Clone, Serialize, Deserialize)]
31pub struct SessionReceipt {
32    /// Always "treeship/session-receipt/v1".
33    #[serde(rename = "type")]
34    pub type_: String,
35
36    /// Schema version. Absent on pre-v0.9.0 receipts (treated as "0").
37    /// Set to "1" for v0.9.0+ receipts.
38    #[serde(default, skip_serializing_if = "Option::is_none")]
39    pub schema_version: Option<String>,
40
41    pub session: SessionSection,
42    pub participants: Participants,
43    pub hosts: Vec<HostInfo>,
44    pub tools: Vec<ToolInfo>,
45    pub agent_graph: AgentGraph,
46    pub timeline: Vec<TimelineEntry>,
47    pub side_effects: SideEffects,
48    pub artifacts: Vec<ArtifactEntry>,
49    pub proofs: ProofsSection,
50    pub merkle: MerkleSection,
51    pub render: RenderConfig,
52    /// Tool usage summary: declared vs actual tools used during the session.
53    #[serde(default, skip_serializing_if = "Option::is_none")]
54    pub tool_usage: Option<ToolUsage>,
55}
56
57/// Tool authorization and usage summary for the session.
58#[derive(Debug, Clone, Default, Serialize, Deserialize)]
59pub struct ToolUsage {
60    /// Tools declared as authorized (from declaration.json).
61    #[serde(default, skip_serializing_if = "Vec::is_empty")]
62    pub declared: Vec<String>,
63    /// Tools actually called during the session with invocation counts.
64    #[serde(default, skip_serializing_if = "Vec::is_empty")]
65    pub actual: Vec<ToolUsageEntry>,
66    /// Tools called that were NOT in the declared list.
67    #[serde(default, skip_serializing_if = "Vec::is_empty")]
68    pub unauthorized: Vec<String>,
69}
70
71/// A single tool's usage count.
72#[derive(Debug, Clone, Serialize, Deserialize)]
73pub struct ToolUsageEntry {
74    pub tool_name: String,
75    pub count: u32,
76}
77
78/// Session metadata section of the receipt.
79#[derive(Debug, Clone, Serialize, Deserialize)]
80pub struct SessionSection {
81    pub id: String,
82    #[serde(skip_serializing_if = "Option::is_none")]
83    pub name: Option<String>,
84    pub mode: LifecycleMode,
85    pub started_at: String,
86    #[serde(skip_serializing_if = "Option::is_none")]
87    pub ended_at: Option<String>,
88    pub status: SessionStatus,
89    #[serde(skip_serializing_if = "Option::is_none")]
90    pub duration_ms: Option<u64>,
91    /// Ship ID this session ran under, parsed from the manifest actor URI
92    /// (`ship://<ship_id>`). Absent on pre-v0.9.0 receipts or when the actor
93    /// URI was not a ship:// URI (e.g. human://alice for a human-led session).
94    /// Cross-verification uses this to check that a receipt and a presented
95    /// Agent Certificate reference the same ship.
96    #[serde(default, skip_serializing_if = "Option::is_none")]
97    pub ship_id: Option<String>,
98    /// Structured narrative for human review. All fields optional.
99    #[serde(default, skip_serializing_if = "Option::is_none")]
100    pub narrative: Option<Narrative>,
101    /// Cumulative input tokens across all agents.
102    #[serde(default)]
103    pub total_tokens_in: u64,
104    /// Cumulative output tokens across all agents.
105    #[serde(default)]
106    pub total_tokens_out: u64,
107}
108
109
110/// Structured narrative for the session summary.
111#[derive(Debug, Clone, Default, Serialize, Deserialize)]
112pub struct Narrative {
113    /// One-line headline: "Verifier refactor completed."
114    #[serde(default, skip_serializing_if = "Option::is_none")]
115    pub headline: Option<String>,
116    /// Multi-sentence summary of what happened.
117    #[serde(default, skip_serializing_if = "Option::is_none")]
118    pub summary: Option<String>,
119    /// What should be reviewed before trusting the output.
120    #[serde(default, skip_serializing_if = "Option::is_none")]
121    pub review: Option<String>,
122}
123
124/// A single timeline entry.
125#[derive(Debug, Clone, Serialize, Deserialize)]
126pub struct TimelineEntry {
127    pub sequence_no: u64,
128    pub timestamp: String,
129    pub event_id: String,
130    pub event_type: String,
131    pub agent_instance_id: String,
132    pub agent_name: String,
133    pub host_id: String,
134    #[serde(skip_serializing_if = "Option::is_none")]
135    pub summary: Option<String>,
136}
137
138/// An artifact referenced in the session.
139#[derive(Debug, Clone, Serialize, Deserialize)]
140pub struct ArtifactEntry {
141    pub artifact_id: String,
142    pub payload_type: String,
143    #[serde(skip_serializing_if = "Option::is_none")]
144    pub digest: Option<String>,
145    #[serde(skip_serializing_if = "Option::is_none")]
146    pub signed_at: Option<String>,
147}
148
149/// Proofs section of the receipt.
150#[derive(Debug, Clone, Default, Serialize, Deserialize)]
151pub struct ProofsSection {
152    #[serde(default)]
153    pub signature_count: u32,
154    #[serde(default)]
155    pub signatures_valid: bool,
156    #[serde(default)]
157    pub merkle_root_valid: bool,
158    #[serde(default)]
159    pub inclusion_proofs_count: u32,
160    #[serde(default)]
161    pub zk_proofs_present: bool,
162    /// Count of events.jsonl lines that were skipped during read_all
163    /// because they failed to deserialize. Set by session::close from
164    /// EventLog::read_all_with_stats. Codex adversarial review finding #8:
165    /// without this in-band signal, a receipt sealed after malformed
166    /// events were silently dropped looks complete to a verifier even
167    /// when it isn't. `treeship package verify` surfaces this as a WARN
168    /// when nonzero. Defaults to 0; absent on pre-v0.9.6 receipts so
169    /// they still verify byte-identical.
170    #[serde(default, skip_serializing_if = "is_zero_u32")]
171    pub event_log_skipped: u32,
172}
173
174fn is_zero_u32(n: &u32) -> bool { *n == 0 }
175
176/// Merkle section of the receipt.
177#[derive(Debug, Clone, Default, Serialize, Deserialize)]
178pub struct MerkleSection {
179    pub leaf_count: usize,
180    #[serde(skip_serializing_if = "Option::is_none")]
181    pub root: Option<String>,
182    #[serde(skip_serializing_if = "Option::is_none")]
183    pub checkpoint_id: Option<String>,
184    #[serde(default, skip_serializing_if = "Vec::is_empty")]
185    pub inclusion_proofs: Vec<InclusionProofEntry>,
186}
187
188/// A Merkle inclusion proof entry.
189#[derive(Debug, Clone, Serialize, Deserialize)]
190pub struct InclusionProofEntry {
191    pub artifact_id: String,
192    pub leaf_index: usize,
193    pub proof: InclusionProof,
194}
195
196// ── Composer ─────────────────────────────────────────────────────────
197
198/// Composes a Session Receipt from events and artifacts.
199pub struct ReceiptComposer;
200
201impl ReceiptComposer {
202    /// Compose a receipt from a session manifest, events, and optional artifact entries.
203    pub fn compose(
204        manifest: &SessionManifest,
205        events: &[SessionEvent],
206        artifact_entries: Vec<ArtifactEntry>,
207    ) -> SessionReceipt {
208        // Build agent graph
209        let agent_graph = AgentGraph::from_events(events);
210
211        // Build side effects
212        let side_effects = SideEffects::from_events(events);
213
214        // Build timeline from all events
215        let mut timeline: Vec<TimelineEntry> = events.iter().map(|e| {
216            TimelineEntry {
217                sequence_no: e.sequence_no,
218                timestamp: e.timestamp.clone(),
219                event_id: e.event_id.clone(),
220                event_type: event_type_label(&e.event_type),
221                agent_instance_id: e.agent_instance_id.clone(),
222                agent_name: e.agent_name.clone(),
223                host_id: e.host_id.clone(),
224                summary: event_summary(&e.event_type),
225            }
226        }).collect();
227
228        // Sort by (timestamp, sequence_no, event_id) for determinism
229        timeline.sort_by(|a, b| {
230            a.timestamp.cmp(&b.timestamp)
231                .then(a.sequence_no.cmp(&b.sequence_no))
232                .then(a.event_id.cmp(&b.event_id))
233        });
234
235        // Compute participants from graph
236        let participants = compute_participants(&agent_graph, manifest);
237
238        // Compute hosts and tools from events
239        let hosts = compute_hosts(events, &manifest.hosts);
240        let tools = compute_tools(events, &manifest.tools);
241
242        // Compute duration from the session close event if present
243        let duration_ms = events.iter().find_map(|e| {
244            if let super::event::EventType::SessionClosed { duration_ms, .. } = &e.event_type {
245                *duration_ms
246            } else {
247                None
248            }
249        });
250
251        // Build Merkle tree from artifact IDs
252        let (merkle_section, merkle_tree) = build_merkle(&artifact_entries);
253
254        // Proofs section. zk_proofs_present defaults to false here;
255        // the CLI caller sets it to true after compose if proof files
256        // exist in the session directory.
257        let proofs = ProofsSection {
258            signature_count: artifact_entries.len() as u32,
259            signatures_valid: true, // Caller should verify
260            merkle_root_valid: merkle_tree.is_some(),
261            inclusion_proofs_count: merkle_section.inclusion_proofs.len() as u32,
262            zk_proofs_present: false,
263            event_log_skipped: 0, // Set by caller after compose (Codex #8)
264        };
265
266        // Compute cost/token totals from agent graph
267        // Cost is deliberately not aggregated. See event.rs comment.
268        let total_tokens_in: u64 = agent_graph.nodes.iter().map(|n| n.tokens_in).sum();
269        let total_tokens_out: u64 = agent_graph.nodes.iter().map(|n| n.tokens_out).sum();
270
271        // Session section
272        let session = SessionSection {
273            id: manifest.session_id.clone(),
274            name: manifest.name.clone(),
275            mode: manifest.mode.clone(),
276            started_at: manifest.started_at.clone(),
277            ended_at: manifest.closed_at.clone(),
278            status: manifest.status.clone(),
279            duration_ms,
280            ship_id: parse_ship_id_from_actor(&manifest.actor),
281            narrative: manifest.summary.as_ref().map(|s| Narrative {
282                headline: manifest.name.clone(),
283                summary: Some(s.clone()),
284                review: None,
285            }),
286            total_tokens_in,
287            total_tokens_out,
288        };
289
290        // Render config
291        let render = RenderConfig {
292            title: manifest.name.clone(),
293            theme: None,
294            sections: RenderConfig::default_sections(),
295            generate_preview: true,
296        };
297
298        // Derive tool usage from side effects + manifest authorized_tools
299        let tool_usage = derive_tool_usage(&side_effects, &manifest.authorized_tools);
300
301        SessionReceipt {
302            type_: RECEIPT_TYPE.into(),
303            schema_version: Some(RECEIPT_SCHEMA_VERSION.into()),
304            session,
305            participants,
306            hosts,
307            tools,
308            agent_graph,
309            timeline,
310            side_effects,
311            artifacts: artifact_entries,
312            proofs,
313            merkle: merkle_section,
314            render,
315            tool_usage,
316        }
317    }
318
319    /// Produce deterministic canonical JSON bytes from a receipt.
320    ///
321    /// Uses serde's field-declaration-order serialization for determinism.
322    /// The resulting bytes are suitable for hashing.
323    pub fn to_canonical_json(receipt: &SessionReceipt) -> Result<Vec<u8>, serde_json::Error> {
324        serde_json::to_vec(receipt)
325    }
326
327    /// Compute SHA-256 digest of the canonical receipt JSON.
328    pub fn digest(receipt: &SessionReceipt) -> Result<String, serde_json::Error> {
329        let bytes = Self::to_canonical_json(receipt)?;
330        let hash = Sha256::digest(&bytes);
331        Ok(format!("sha256:{}", hex::encode(hash)))
332    }
333}
334
335// ── Helpers ──────────────────────────────────────────────────────────
336
337fn compute_participants(graph: &AgentGraph, manifest: &SessionManifest) -> Participants {
338    use std::collections::BTreeSet;
339
340    let mut tool_runtimes: BTreeSet<String> = BTreeSet::new();
341    // Count unique agents
342    let total_agents = graph.nodes.len() as u32;
343    let spawned_subagents = graph.spawn_count();
344    let handoffs = graph.handoff_count();
345    let max_depth = graph.max_depth();
346    let host_ids = graph.host_ids();
347
348    // Collect tool runtimes from events in manifest
349    for tool in &manifest.tools {
350        if let Some(ref rt) = tool.tool_runtime_id {
351            tool_runtimes.insert(rt.clone());
352        }
353    }
354
355    // Find root agent (depth 0, first started)
356    let root = graph.nodes.iter()
357        .filter(|n| n.depth == 0)
358        .min_by_key(|n| n.started_at.as_deref().unwrap_or(""))
359        .map(|n| n.agent_instance_id.clone());
360
361    // Find final output agent (last completed at max depth or last completed overall)
362    let final_output = graph.nodes.iter()
363        .filter(|n| n.completed_at.is_some())
364        .max_by_key(|n| n.completed_at.as_deref().unwrap_or(""))
365        .map(|n| n.agent_instance_id.clone());
366
367    Participants {
368        root_agent_instance_id: root.or(manifest.participants.root_agent_instance_id.clone()),
369        final_output_agent_instance_id: final_output.or(manifest.participants.final_output_agent_instance_id.clone()),
370        total_agents,
371        spawned_subagents,
372        handoffs,
373        max_depth,
374        hosts: host_ids.len() as u32,
375        tool_runtimes: tool_runtimes.len() as u32,
376    }
377}
378
379fn compute_hosts(events: &[SessionEvent], manifest_hosts: &[HostInfo]) -> Vec<HostInfo> {
380    use std::collections::BTreeMap;
381
382    let mut hosts: BTreeMap<String, HostInfo> = BTreeMap::new();
383
384    // Seed from manifest
385    for h in manifest_hosts {
386        hosts.insert(h.host_id.clone(), h.clone());
387    }
388
389    // Discover from events
390    for e in events {
391        hosts.entry(e.host_id.clone()).or_insert_with(|| HostInfo {
392            host_id: e.host_id.clone(),
393            hostname: None,
394            os: None,
395            arch: None,
396        });
397    }
398
399    hosts.into_values().collect()
400}
401
402fn compute_tools(events: &[SessionEvent], manifest_tools: &[ToolInfo]) -> Vec<ToolInfo> {
403    use std::collections::BTreeMap;
404
405    let mut tools: BTreeMap<String, ToolInfo> = BTreeMap::new();
406
407    // Seed from manifest
408    for t in manifest_tools {
409        tools.insert(t.tool_id.clone(), t.clone());
410    }
411
412    // Count tool invocations from events
413    for e in events {
414        if let super::event::EventType::AgentCalledTool { ref tool_name, .. } = e.event_type {
415            let entry = tools.entry(tool_name.clone()).or_insert_with(|| ToolInfo {
416                tool_id: tool_name.clone(),
417                tool_name: tool_name.clone(),
418                tool_runtime_id: e.tool_runtime_id.clone(),
419                invocation_count: 0,
420            });
421            entry.invocation_count += 1;
422        }
423    }
424
425    tools.into_values().collect()
426}
427
428fn build_merkle(artifacts: &[ArtifactEntry]) -> (MerkleSection, Option<MerkleTree>) {
429    if artifacts.is_empty() {
430        return (MerkleSection::default(), None);
431    }
432
433    let mut tree = MerkleTree::new();
434    for art in artifacts {
435        tree.append(&art.artifact_id);
436    }
437
438    let root = tree.root().map(|r| format!("mroot_{}", hex::encode(r)));
439
440    // Build inclusion proofs for each artifact
441    let inclusion_proofs: Vec<InclusionProofEntry> = artifacts.iter().enumerate()
442        .filter_map(|(i, art)| {
443            tree.inclusion_proof(i).map(|proof| InclusionProofEntry {
444                artifact_id: art.artifact_id.clone(),
445                leaf_index: i,
446                proof,
447            })
448        })
449        .collect();
450
451    let section = MerkleSection {
452        leaf_count: artifacts.len(),
453        root,
454        checkpoint_id: None,
455        inclusion_proofs,
456    };
457
458    (section, Some(tree))
459}
460
461/// Extract the ship_id from an actor URI of the form `ship://<id>`.
462/// Returns None for other URI schemes (human://, agent://) or malformed values.
463pub fn parse_ship_id_from_actor(actor: &str) -> Option<String> {
464    let rest = actor.strip_prefix("ship://")?;
465    // Strip any trailing path segment so `ship://ship_abc/foo` -> `ship_abc`.
466    let id = rest.split('/').next().unwrap_or(rest);
467    if id.is_empty() { None } else { Some(id.to_string()) }
468}
469
470/// Extract a human-readable label from an EventType.
471/// Derive tool usage from side effects and the declared authorized tools list.
472///
473/// Bug Codex caught in adversarial review: previously this function counted
474/// only `side_effects.tool_invocations` (built from `EventType::AgentCalledTool`).
475/// But Claude Code's PostToolUse hook emits SPECIALIZED events for built-in
476/// tools (`agent.wrote_file` for Write/Edit, `agent.completed_process` for
477/// Bash, `agent.read_file` for Read, etc) -- those events never landed in
478/// `tool_invocations`, so a certificate that omitted "Bash" or "Write"
479/// passed cross-verification cleanly even when the agent ran them.
480///
481/// The fix: also count side effects from specialized event types under
482/// canonical tool names that match what an operator would declare in
483/// `bounded_actions`. Naming follows Claude Code conventions (Read, Write,
484/// Bash, WebFetch) since those are the tools users actually declare. A
485/// cert that uses an alternate naming scheme (e.g. `files.write`) needs
486/// to declare both for now -- a future TODO is canonical mapping at the
487/// cert layer.
488/// Side-effect canonical mapping for tool authorization.
489///
490/// Each entry maps a side-effect bucket to a canonical tool name AND a
491/// list of accepted aliases. The canonical name is what gets recorded
492/// in `tool_usage.actual`. Any alias from the authorized_tools list
493/// counts as authorization for the canonical name.
494///
495/// Codex round-2 caught two bugs in the round-1 fix:
496///
497/// 1. The round-1 mapping used Claude-Code TitleCase ("Read", "Write",
498///    "Bash") but the existing CLI -- `treeship declare --tools
499///    read_file,write_file,bash` per declare.rs:80 and `treeship agent
500///    register --tools read_file,write_file,bash` per main.rs:226 --
501///    teaches users lowercase snake_case names. So a cert that follows
502///    the documented convention got every actual tool flagged as
503///    unauthorized. Aliases close that gap: declarations in either
504///    convention authorize the same canonical entry.
505///
506/// 2. The round-1 logic counted side effects regardless of provenance.
507///    `git-reconcile` synthetic writes (the backstop layer) registered
508///    as tool use even though no actual tool was directly attributed
509///    for them. A build script that touched a file made the receipt
510///    say "Write tool was used", and the cert had to authorize Write
511///    or fail cross-verify -- even though the agent never invoked any
512///    Write tool. Below, only direct-attribution sources (`hook`,
513///    `mcp`, `shell-wrap`, `session-event-cli`, and untagged legacy
514///    events) count toward tool usage. Backstop sources (`git-reconcile`,
515///    `daemon-atime`) surface in the receipt's "Files changed" section
516///    so the reader sees the change, but they do NOT claim that an
517///    agent tool was the proximate cause. See source_attributes_a_tool
518///    below for the authoritative allow list.
519const TOOL_ALIASES: &[(&str, &[&str])] = &[
520    // Canonical first; rest are accepted aliases.
521    ("read_file",  &["read_file", "Read"]),
522    ("write_file", &["write_file", "Write", "Edit", "MultiEdit", "NotebookEdit", "edit_file"]),
523    ("bash",       &["bash", "Bash", "shell"]),
524    ("web_fetch",  &["web_fetch", "WebFetch", "webfetch"]),
525];
526
527/// Returns true iff `source` represents a direct tool attribution that
528/// should count toward `tool_usage.actual`.
529///
530/// Direct attribution sources -- a real tool fired and the channel
531/// captured it:
532///   - `hook`              integration hook saw the tool fire
533///   - `mcp`               promoted from MCP-bridge agent.called_tool
534///   - `shell-wrap`        `treeship wrap` captured a shell command
535///   - `session-event-cli` `treeship session event` from a hook script.
536///                         The Claude Code plugin's PostToolUse hook
537///                         calls `treeship session event --type
538///                         agent.wrote_file --file X`, and the CLI
539///                         tags those as "session-event-cli" -- so
540///                         excluding this label would make every
541///                         claude-code-plugin event invisible to
542///                         cross-verify.
543///   - None                legacy untagged event (back-compat)
544///
545/// Backstop / inference sources -- a file changed but no tool was
546/// directly attributed. Surface in the receipt's "Files changed"
547/// section so the reader sees the change but they must NOT inflate
548/// tool_usage:
549///   - `git-reconcile`     git diff at session close
550///   - `daemon-atime`      atime-based file detection
551fn source_attributes_a_tool(source: Option<&str>) -> bool {
552    matches!(
553        source,
554        None
555            | Some("hook")
556            | Some("mcp")
557            | Some("shell-wrap")
558            | Some("session-event-cli"),
559    )
560}
561
562/// Counts side effects by canonical tool name, filtering out
563/// non-attribution sources (git-reconcile, daemon-atime).
564fn count_attributed<'a, F>(
565    items: usize,
566    source_at: F,
567    canonical: &str,
568    counts: &mut std::collections::BTreeMap<String, u32>,
569)
570where
571    F: Fn(usize) -> Option<&'a str>,
572{
573    let n: u32 = (0..items)
574        .filter(|i| source_attributes_a_tool(source_at(*i)))
575        .count() as u32;
576    if n > 0 {
577        *counts.entry(canonical.to_string()).or_insert(0) += n;
578    }
579}
580
581fn derive_tool_usage(
582    side_effects: &SideEffects,
583    authorized_tools: &[String],
584) -> Option<ToolUsage> {
585    use std::collections::BTreeMap;
586
587    let total_specialized = side_effects.files_read.len()
588        + side_effects.files_written.len()
589        + side_effects.processes.len()
590        + side_effects.network_connections.len();
591
592    if side_effects.tool_invocations.is_empty()
593        && total_specialized == 0
594        && authorized_tools.is_empty()
595    {
596        return None;
597    }
598
599    let mut counts: BTreeMap<String, u32> = BTreeMap::new();
600
601    // Generic agent.called_tool events use the tool's actual name.
602    // The MCP bridge writes meta.source = "mcp-bridge" (which is not
603    // in source_attributes_a_tool's allow list) but tool_invocations
604    // come ONLY from agent.called_tool, which is direct attribution
605    // by definition -- so count all of them, no source filter applies
606    // here. (The bridge tool name is the source.)
607    for inv in &side_effects.tool_invocations {
608        *counts.entry(inv.tool_name.clone()).or_insert(0) += 1;
609    }
610
611    // Specialized side effects, source-filtered: only direct
612    // attribution (hook / mcp / shell-wrap / untagged-legacy) counts.
613    // git-reconcile and friends surface in the "Files changed" section
614    // for the reader but do NOT inflate tool_usage.
615    let fr = &side_effects.files_read;
616    count_attributed(
617        fr.len(),
618        |i| fr[i].source.as_deref(),
619        "read_file",
620        &mut counts,
621    );
622    let fw = &side_effects.files_written;
623    count_attributed(
624        fw.len(),
625        |i| fw[i].source.as_deref(),
626        "write_file",
627        &mut counts,
628    );
629    let pr = &side_effects.processes;
630    count_attributed(
631        pr.len(),
632        |i| pr[i].source.as_deref(),
633        "bash",
634        &mut counts,
635    );
636    // network_connections has no source field today; treat all as
637    // attributed (this matches the round-1 behavior since there's no
638    // backstop layer producing network entries).
639    if !side_effects.network_connections.is_empty() {
640        *counts.entry("web_fetch".to_string()).or_insert(0) +=
641            side_effects.network_connections.len() as u32;
642    }
643
644    let actual: Vec<ToolUsageEntry> = counts.iter()
645        .map(|(name, &count)| ToolUsageEntry { tool_name: name.clone(), count })
646        .collect();
647
648    // Authorization check uses alias resolution: an actual tool is
649    // unauthorized only if NONE of its aliases are in the declared
650    // list. So a declaration of "read_file" authorizes both "Read"
651    // (Claude convention) and "read_file" (CLI convention) when they
652    // produce the canonical "read_file" actual entry.
653    let unauthorized = if authorized_tools.is_empty() {
654        Vec::new()
655    } else {
656        let declared_set: std::collections::BTreeSet<&str> = authorized_tools.iter()
657            .map(|s| s.as_str())
658            .collect();
659        counts.keys()
660            .filter(|actual_name| !is_authorized(actual_name, &declared_set))
661            .cloned()
662            .collect()
663    };
664
665    Some(ToolUsage {
666        declared: authorized_tools.to_vec(),
667        actual,
668        unauthorized,
669    })
670}
671
672/// Returns true if `actual_name` (or any of its declared aliases) is
673/// in the declared set. Aliases mean a cert can use either Claude
674/// convention or snake_case CLI convention and still authorize the
675/// same canonical bucket.
676fn is_authorized(actual_name: &str, declared_set: &std::collections::BTreeSet<&str>) -> bool {
677    // Direct hit: the declared set names this tool exactly.
678    if declared_set.contains(actual_name) {
679        return true;
680    }
681    // Alias hit: walk the canonical mapping and see if any alias of
682    // the canonical bucket the actual_name belongs to is in declared.
683    for (canonical, aliases) in TOOL_ALIASES {
684        if *canonical == actual_name || aliases.contains(&actual_name) {
685            for alias in *aliases {
686                if declared_set.contains(*alias) {
687                    return true;
688                }
689            }
690            return false;
691        }
692    }
693    false
694}
695
696fn event_type_label(et: &super::event::EventType) -> String {
697    use super::event::EventType::*;
698    match et {
699        SessionStarted => "session.started",
700        SessionClosed { .. } => "session.closed",
701        AgentStarted { .. } => "agent.started",
702        AgentSpawned { .. } => "agent.spawned",
703        AgentHandoff { .. } => "agent.handoff",
704        AgentCollaborated { .. } => "agent.collaborated",
705        AgentReturned { .. } => "agent.returned",
706        AgentCompleted { .. } => "agent.completed",
707        AgentFailed { .. } => "agent.failed",
708        AgentCalledTool { .. } => "agent.called_tool",
709        AgentReadFile { .. } => "agent.read_file",
710        AgentWroteFile { .. } => "agent.wrote_file",
711        AgentOpenedPort { .. } => "agent.opened_port",
712        AgentConnectedNetwork { .. } => "agent.connected_network",
713        AgentStartedProcess { .. } => "agent.started_process",
714        AgentCompletedProcess { .. } => "agent.completed_process",
715        AgentDecision { .. } => "agent.decision",
716    }.into()
717}
718
719/// Optional human-readable summary from an EventType.
720fn event_summary(et: &super::event::EventType) -> Option<String> {
721    use super::event::EventType::*;
722    match et {
723        SessionStarted => Some("Session started".into()),
724        SessionClosed { summary, .. } => summary.clone().or(Some("Session closed".into())),
725        AgentSpawned { reason, .. } => reason.clone(),
726        AgentHandoff { from_agent_instance_id, to_agent_instance_id, .. } => {
727            Some(format!("{from_agent_instance_id} -> {to_agent_instance_id}"))
728        }
729        AgentCalledTool { tool_name, .. } => Some(format!("Called {tool_name}")),
730        AgentReadFile { file_path, .. } => Some(format!("Read {file_path}")),
731        AgentWroteFile { file_path, .. } => Some(format!("Wrote {file_path}")),
732        AgentOpenedPort { port, .. } => Some(format!("Opened port {port}")),
733        AgentConnectedNetwork { destination, .. } => Some(format!("Connected to {destination}")),
734        AgentStartedProcess { process_name, .. } => Some(format!("Started {process_name}")),
735        AgentCompletedProcess { process_name, exit_code, .. } => {
736            Some(format!("Completed {process_name} (exit {})", exit_code.unwrap_or(-1)))
737        }
738        AgentCompleted { termination_reason } => termination_reason.clone().or(Some("Agent completed".into())),
739        AgentFailed { reason } => reason.clone().or(Some("Agent failed".into())),
740        AgentDecision { model, summary, provider, .. } => {
741            let mut parts = Vec::new();
742            if let Some(s) = summary { parts.push(s.clone()); }
743            if let Some(m) = model { parts.push(format!("model: {m}")); }
744            if let Some(p) = provider { parts.push(format!("via {p}")); }
745            if parts.is_empty() { Some("LLM decision".into()) } else { Some(parts.join(" | ")) }
746        }
747        _ => None,
748    }
749}
750
751#[cfg(test)]
752mod tests {
753    use super::*;
754    use crate::session::event::*;
755
756    fn make_manifest() -> SessionManifest {
757        SessionManifest::new(
758            "ssn_001".into(),
759            "agent://test".into(),
760            "2026-04-05T08:00:00Z".into(),
761            1743843600000,
762        )
763    }
764
765    /// Module-level event constructor so the tool-authorization regression
766    /// tests below can reuse it without each redefining the closure.
767    fn mk(seq: u64, inst: &str, et: EventType) -> SessionEvent {
768        SessionEvent {
769            session_id: "ssn_001".into(),
770            event_id: format!("evt_{:016x}", seq),
771            timestamp: format!("2026-04-05T08:{:02}:00Z", seq),
772            sequence_no: seq,
773            trace_id: "trace_1".into(),
774            span_id: format!("span_{seq}"),
775            parent_span_id: None,
776            agent_id: format!("agent://{inst}"),
777            agent_instance_id: inst.into(),
778            agent_name: inst.into(),
779            agent_role: None,
780            host_id: "host_1".into(),
781            tool_runtime_id: None,
782            event_type: et,
783            artifact_ref: None,
784            meta: None,
785        }
786    }
787
788    fn make_events() -> Vec<SessionEvent> {
789        vec![
790            mk(0, "root", EventType::SessionStarted),
791            mk(1, "root", EventType::AgentStarted { parent_agent_instance_id: None }),
792            mk(2, "worker", EventType::AgentSpawned { spawned_by_agent_instance_id: "root".into(), reason: Some("review".into()) }),
793            mk(3, "worker", EventType::AgentCalledTool { tool_name: "read_file".into(), tool_input_digest: None, tool_output_digest: None, duration_ms: Some(5) }),
794            mk(4, "worker", EventType::AgentWroteFile { file_path: "src/fix.rs".into(), digest: None, operation: None, additions: None, deletions: None }),
795            mk(5, "worker", EventType::AgentCompleted { termination_reason: None }),
796            mk(6, "root", EventType::SessionClosed { summary: Some("Done".into()), duration_ms: Some(360000) }),
797        ]
798    }
799
800    #[test]
801    fn compose_receipt() {
802        let manifest = make_manifest();
803        let events = make_events();
804        let artifacts = vec![
805            ArtifactEntry { artifact_id: "art_001".into(), payload_type: "action".into(), digest: None, signed_at: None },
806            ArtifactEntry { artifact_id: "art_002".into(), payload_type: "action".into(), digest: None, signed_at: None },
807        ];
808
809        let receipt = ReceiptComposer::compose(&manifest, &events, artifacts);
810
811        assert_eq!(receipt.type_, RECEIPT_TYPE);
812        assert_eq!(receipt.session.id, "ssn_001");
813        assert_eq!(receipt.timeline.len(), 7);
814        assert_eq!(receipt.agent_graph.nodes.len(), 2); // root + worker
815        assert_eq!(receipt.side_effects.files_written.len(), 1);
816        assert_eq!(receipt.merkle.leaf_count, 2);
817        assert!(receipt.merkle.root.is_some());
818    }
819
820    #[test]
821    fn new_receipts_carry_schema_version() {
822        let manifest = make_manifest();
823        let events = make_events();
824        let artifacts = vec![
825            ArtifactEntry { artifact_id: "art_001".into(), payload_type: "action".into(), digest: None, signed_at: None },
826        ];
827        let receipt = ReceiptComposer::compose(&manifest, &events, artifacts);
828        assert_eq!(receipt.schema_version.as_deref(), Some(RECEIPT_SCHEMA_VERSION));
829        // And it shows up in canonical JSON.
830        let json = String::from_utf8(ReceiptComposer::to_canonical_json(&receipt).unwrap()).unwrap();
831        assert!(json.contains(r#""schema_version":"1""#), "missing schema_version: {json}");
832    }
833
834    #[test]
835    fn legacy_receipt_without_schema_version_round_trips_byte_identical() {
836        // Simulate a pre-v0.9.0 receipt by composing one and stripping the
837        // schema_version field. Re-serializing must produce byte-identical
838        // output so the package-level determinism check keeps passing for
839        // old receipts that nobody can re-sign.
840        let manifest = make_manifest();
841        let events = make_events();
842        let artifacts = vec![
843            ArtifactEntry { artifact_id: "art_001".into(), payload_type: "action".into(), digest: None, signed_at: None },
844        ];
845        let mut receipt = ReceiptComposer::compose(&manifest, &events, artifacts);
846        receipt.schema_version = None; // mimic a legacy receipt
847
848        let original = ReceiptComposer::to_canonical_json(&receipt).unwrap();
849        // Verify the field is omitted, not serialized as null.
850        let original_str = std::str::from_utf8(&original).unwrap();
851        assert!(!original_str.contains("schema_version"),
852            "schema_version must be skipped when None");
853
854        let parsed: SessionReceipt = serde_json::from_slice(&original).unwrap();
855        assert!(parsed.schema_version.is_none(), "legacy receipts must parse with schema_version=None");
856
857        let reserialized = ReceiptComposer::to_canonical_json(&parsed).unwrap();
858        assert_eq!(original, reserialized,
859            "legacy receipt must round-trip byte-identical so package determinism check passes");
860    }
861
862    #[test]
863    fn canonical_json_is_deterministic() {
864        let manifest = make_manifest();
865        let events = make_events();
866        let artifacts = vec![
867            ArtifactEntry { artifact_id: "art_001".into(), payload_type: "action".into(), digest: None, signed_at: None },
868        ];
869
870        let r1 = ReceiptComposer::compose(&manifest, &events, artifacts.clone());
871        let r2 = ReceiptComposer::compose(&manifest, &events, artifacts);
872
873        let j1 = ReceiptComposer::to_canonical_json(&r1).unwrap();
874        let j2 = ReceiptComposer::to_canonical_json(&r2).unwrap();
875        assert_eq!(j1, j2);
876
877        let d1 = ReceiptComposer::digest(&r1).unwrap();
878        let d2 = ReceiptComposer::digest(&r2).unwrap();
879        assert_eq!(d1, d2);
880    }
881
882    // ── Tool authorization regression tests (Codex finding #1) ──
883    //
884    // Specialized event types (agent.wrote_file, agent.completed_process,
885    // agent.read_file) must contribute to tool_usage.actual so that a
886    // certificate's bounded_actions list can correctly flag unauthorized
887    // built-in tool usage. Before this fix, only agent.called_tool fed
888    // tool_usage.actual, so a cert that omitted "Bash" still passed even
889    // when the agent ran Bash via Claude Code's built-in.
890
891    fn manifest_with_authorized(tools: Vec<&str>) -> SessionManifest {
892        let mut m = make_manifest();
893        m.authorized_tools = tools.into_iter().map(String::from).collect();
894        m
895    }
896
897    #[test]
898    fn cert_omitting_bash_flags_unauthorized_when_session_runs_bash() {
899        // Cert uses CLI-documented snake_case names (declare.rs:80,
900        // main.rs:226). Round-2 fix: canonical actual is "bash" not
901        // "Bash"; round-1 was flagging mismatches the wrong way.
902        let manifest = manifest_with_authorized(vec!["read_file", "write_file"]); // NO bash
903        let events = vec![
904            mk(0, "root", EventType::SessionStarted),
905            mk(1, "agent", EventType::AgentCompletedProcess {
906                process_name: "rm -rf /".into(),
907                exit_code: Some(0),
908                duration_ms: Some(50),
909                command: Some("rm -rf /".into()),
910            }),
911            mk(2, "root", EventType::SessionClosed { summary: None, duration_ms: Some(1000) }),
912        ];
913        let receipt = ReceiptComposer::compose(&manifest, &events, vec![]);
914        let tu = receipt.tool_usage.expect("tool_usage must be populated");
915        assert!(
916            tu.unauthorized.iter().any(|t| t == "bash"),
917            "bash must be flagged as unauthorized when cert omits it; got unauthorized={:?}, actual={:?}",
918            tu.unauthorized, tu.actual,
919        );
920    }
921
922    #[test]
923    fn cert_omitting_write_flags_unauthorized_when_session_writes_file() {
924        let manifest = manifest_with_authorized(vec!["read_file", "bash"]); // NO write_file
925        let events = vec![
926            mk(0, "root", EventType::SessionStarted),
927            mk(1, "agent", EventType::AgentWroteFile {
928                file_path: "src/secret.rs".into(),
929                digest: None, operation: Some("modified".into()),
930                additions: Some(10), deletions: Some(0),
931            }),
932            mk(2, "root", EventType::SessionClosed { summary: None, duration_ms: Some(1000) }),
933        ];
934        let receipt = ReceiptComposer::compose(&manifest, &events, vec![]);
935        let tu = receipt.tool_usage.expect("tool_usage must be populated");
936        assert!(
937            tu.unauthorized.iter().any(|t| t == "write_file"),
938            "write_file must be flagged as unauthorized when cert omits it; got unauthorized={:?}, actual={:?}",
939            tu.unauthorized, tu.actual,
940        );
941    }
942
943    #[test]
944    fn cert_includes_read_write_bash_passes_clean_when_all_used() {
945        let manifest = manifest_with_authorized(vec!["read_file", "write_file", "bash"]);
946        let events = vec![
947            mk(0, "root", EventType::SessionStarted),
948            mk(1, "agent", EventType::AgentReadFile { file_path: "package.json".into(), digest: None }),
949            mk(2, "agent", EventType::AgentWroteFile {
950                file_path: "src/lib.rs".into(),
951                digest: None, operation: Some("modified".into()),
952                additions: Some(5), deletions: Some(2),
953            }),
954            mk(3, "agent", EventType::AgentCompletedProcess {
955                process_name: "bun test".into(),
956                exit_code: Some(0), duration_ms: Some(2000),
957                command: Some("bun test".into()),
958            }),
959            mk(4, "root", EventType::SessionClosed { summary: None, duration_ms: Some(5000) }),
960        ];
961        let receipt = ReceiptComposer::compose(&manifest, &events, vec![]);
962        let tu = receipt.tool_usage.expect("tool_usage must be populated");
963        assert!(
964            tu.unauthorized.is_empty(),
965            "all tools declared in cert should pass clean; got unauthorized={:?}",
966            tu.unauthorized,
967        );
968        // The actual list uses canonical lowercase names that match what
969        // `treeship declare --tools` and `treeship agent register --tools`
970        // teach (declare.rs:80, main.rs:226).
971        let actual_names: std::collections::BTreeSet<String> =
972            tu.actual.iter().map(|e| e.tool_name.clone()).collect();
973        assert!(actual_names.contains("read_file"));
974        assert!(actual_names.contains("write_file"));
975        assert!(actual_names.contains("bash"));
976    }
977
978    #[test]
979    fn webfetch_unauthorized_flagged_when_cert_omits_it() {
980        let manifest = manifest_with_authorized(vec!["read_file", "write_file", "bash"]); // NO web_fetch
981        let events = vec![
982            mk(0, "root", EventType::SessionStarted),
983            mk(1, "agent", EventType::AgentConnectedNetwork {
984                destination: "evil.example.com".into(),
985                port: Some(443),
986            }),
987            mk(2, "root", EventType::SessionClosed { summary: None, duration_ms: Some(1000) }),
988        ];
989        let receipt = ReceiptComposer::compose(&manifest, &events, vec![]);
990        let tu = receipt.tool_usage.expect("tool_usage must be populated");
991        assert!(
992            tu.unauthorized.iter().any(|t| t == "web_fetch"),
993            "web_fetch must be flagged as unauthorized when cert omits it; got unauthorized={:?}",
994            tu.unauthorized,
995        );
996    }
997
998    // ── Round-2 fix tests: alias matching + source filtering ──
999
1000    fn evt_with_source(event_type: EventType, source: &str) -> SessionEvent {
1001        let mut e = mk(99, "agent", event_type);
1002        e.meta = Some(serde_json::json!({"source": source}));
1003        e
1004    }
1005
1006    #[test]
1007    fn titlecase_cert_authorizes_canonical_snake_actuals_via_alias() {
1008        // Operator declares Claude convention. Aliases map "Read" to
1009        // canonical "read_file", "Write" to "write_file", etc.
1010        let manifest = manifest_with_authorized(vec!["Read", "Write", "Bash"]);
1011        let events = vec![
1012            mk(0, "root", EventType::SessionStarted),
1013            mk(1, "agent", EventType::AgentReadFile { file_path: "x".into(), digest: None }),
1014            mk(2, "agent", EventType::AgentWroteFile {
1015                file_path: "y".into(),
1016                digest: None, operation: None, additions: None, deletions: None,
1017            }),
1018            mk(3, "agent", EventType::AgentCompletedProcess {
1019                process_name: "z".into(),
1020                exit_code: Some(0), duration_ms: Some(1), command: None,
1021            }),
1022            mk(4, "root", EventType::SessionClosed { summary: None, duration_ms: Some(1000) }),
1023        ];
1024        let tu = ReceiptComposer::compose(&manifest, &events, vec![]).tool_usage.unwrap();
1025        assert!(
1026            tu.unauthorized.is_empty(),
1027            "TitleCase declarations must authorize canonical snake_case actuals via aliases; \
1028             got unauthorized={:?}",
1029            tu.unauthorized,
1030        );
1031    }
1032
1033    #[test]
1034    fn edit_alias_authorizes_specialized_wrote_file() {
1035        // Operator declares "Edit" specifically. post-tool-use.sh
1036        // emits agent.wrote_file for Edit/MultiEdit alike, so the
1037        // canonical actual is "write_file". Edit is in the write_file
1038        // alias list, so the cert authorizes.
1039        let manifest = manifest_with_authorized(vec!["Edit"]);
1040        let events = vec![
1041            mk(0, "root", EventType::SessionStarted),
1042            mk(1, "agent", EventType::AgentWroteFile {
1043                file_path: "x".into(),
1044                digest: None, operation: None, additions: None, deletions: None,
1045            }),
1046            mk(2, "root", EventType::SessionClosed { summary: None, duration_ms: Some(1000) }),
1047        ];
1048        let tu = ReceiptComposer::compose(&manifest, &events, vec![]).tool_usage.unwrap();
1049        assert!(tu.unauthorized.is_empty(), "Edit alias must authorize write_file");
1050    }
1051
1052    #[test]
1053    fn git_reconcile_writes_dont_count_toward_tool_usage() {
1054        // Backstop evidence -- not direct tool attribution.
1055        // A git-reconciled change must NOT make the cert require
1056        // write_file authorization, because no Write tool was invoked.
1057        let manifest = manifest_with_authorized(vec!["read_file"]);
1058        let events = vec![
1059            mk(0, "root", EventType::SessionStarted),
1060            evt_with_source(
1061                EventType::AgentWroteFile {
1062                    file_path: "CHANGELOG.md".into(),
1063                    digest: None, operation: Some("modified".into()),
1064                    additions: Some(7), deletions: Some(2),
1065                },
1066                "git-reconcile",
1067            ),
1068            mk(2, "root", EventType::SessionClosed { summary: None, duration_ms: Some(1000) }),
1069        ];
1070        let tu = ReceiptComposer::compose(&manifest, &events, vec![]).tool_usage.unwrap();
1071        assert!(
1072            !tu.unauthorized.iter().any(|t| t == "write_file"),
1073            "git-reconcile entries must NOT count toward tool_usage; \
1074             got unauthorized={:?}, actual={:?}",
1075            tu.unauthorized, tu.actual,
1076        );
1077        let actual_names: std::collections::BTreeSet<String> =
1078            tu.actual.iter().map(|e| e.tool_name.clone()).collect();
1079        assert!(!actual_names.contains("write_file"),
1080            "actual must not include backstop-only writes");
1081    }
1082
1083    // session-event-cli is a direct-attribution source -- the standard
1084    // label the CLI stamps on events emitted by claude-code-plugin's
1085    // PostToolUse hook. So it counts toward tool_usage.actual just like
1086    // hook/mcp/shell-wrap do. The end-to-end test for this lives in
1087    // the targeted acceptance suite (T1) rather than as a unit test
1088    // here, because it requires the full event-emission + receipt-
1089    // composition pipeline running through `treeship session event`.
1090
1091    #[test]
1092    fn hook_emitted_writes_still_count_toward_tool_usage() {
1093        // Positive case: regular hook-emitted write IS direct attribution.
1094        let manifest = manifest_with_authorized(vec!["read_file"]); // NO write_file
1095        let events = vec![
1096            mk(0, "root", EventType::SessionStarted),
1097            evt_with_source(
1098                EventType::AgentWroteFile {
1099                    file_path: "src/x.rs".into(),
1100                    digest: None, operation: None, additions: None, deletions: None,
1101                },
1102                "hook",
1103            ),
1104            mk(2, "root", EventType::SessionClosed { summary: None, duration_ms: Some(1000) }),
1105        ];
1106        let tu = ReceiptComposer::compose(&manifest, &events, vec![]).tool_usage.unwrap();
1107        assert!(
1108            tu.unauthorized.iter().any(|t| t == "write_file"),
1109            "hook-emitted writes MUST count toward tool_usage; got unauthorized={:?}",
1110            tu.unauthorized,
1111        );
1112    }
1113
1114    #[test]
1115    fn legacy_untagged_writes_count_for_back_compat() {
1116        // Pre-v0.9.6 events have no source tag. Treat as attributed
1117        // (back-compat: receipts produced before source labeling existed).
1118        let manifest = manifest_with_authorized(vec!["read_file"]); // NO write_file
1119        let events = vec![
1120            mk(0, "root", EventType::SessionStarted),
1121            mk(1, "agent", EventType::AgentWroteFile {
1122                file_path: "x".into(),
1123                digest: None, operation: None, additions: None, deletions: None,
1124            }),
1125            mk(2, "root", EventType::SessionClosed { summary: None, duration_ms: Some(1000) }),
1126        ];
1127        let tu = ReceiptComposer::compose(&manifest, &events, vec![]).tool_usage.unwrap();
1128        assert!(
1129            tu.unauthorized.iter().any(|t| t == "write_file"),
1130            "legacy untagged writes must count for back-compat",
1131        );
1132    }
1133}